Due to the particularity and complexity of English phonetic learning, English phonetic researches have received less attention compared with other language skills. The emergence of computer-assisted language learning makes speech visualization possible. However, relevant studies have shown that although visual feedback has a positive impact on speech learning, most studies only focus on the super-segmental features of speech sound, while other parts of the speech systems are not involved. Based on the use of Praat, a speech visual analysis software, this study uses visual feedback paradigm (VFP) in speech teaching, and takes the speech intensity ratio as the test standard to study and analyze the situation of learners' speech learning, and makes the following conclusions:first, using VFP in speech teaching could achieve better results than traditional auditory training, and the speech improvement achieved has continuous effect; second, the impact of VFP at the word level could be extended to similar utterance or sentence level, and the promotion is also persistent; third, students make positive comments towards VFP. They believe that visual feedback provides obvious image differences and can reflect the relative length and intensity of individual phonemes, thus providing guidance for their pronunciation. This study demonstrates the effectiveness and generalization of VFP for English pronunciation learning, as well as the potential practical value of using VFP in second language speech classes.